Usage of the rnashape web server

 

The rnashape server processes un-normalized SHAPE data values and computes a Bayesian posterior probability of base pairing for each residue. This is achieved using probabilistic models that were generated using SHAPE data values (using NMIA as reagent; gels were quantified using the Safa software) of RNAs with known 3D structure.

 

Input Format

The input format that the web server expects is best explained using an example. Listed below are SHAPE experiment values corresponding to the group II self-splicing intron (PDB: 1KXK). The shown values correspond to 13.5mM NMIA concentration for the column “Value” and 0mM NMIA concentration for the column control. Also shown is the sequence in the first column and the residue ids corresponding to the PDB structure for which experimental data is available.

 

The example data (NOT in the input format) looks like:

Seq   Id    Value Control

C                   44               27500  10400

G                   45               27500  10100

U                   46               16800  18100

A                   47               10500  9220

C                   48               10700  14900

G                   49               7880       8960

G                   50               11000  10700

U                   51               13300  13900

U                   52               NA              NA

C                   53               14100  15100

U                   54               35300  46400

A                   55               26200  14600

U                   56               28300  15600

G                   57               10500  6930

C                   58               10200  9550

C                   59               17400  17700

C                   60               18400  13400

G                   61               15600  11400

G                   62               14700  12200

G                   63               18300  24100

G                   64               21400  12400

G                   65               16900  20200

A                   66               28500  14200

A                   67               38900  25800

A                   68               44800  20700

A                   69               42000  32300

 

 

The user should submit the difference between the SHAPE experiment data values and the control data values, one value per line. So the text area in the submission form should for this example be filled with the following set of values:

17100

17400

-1300

1280

-4200

-1080

300

-600

NA

-1000

-11100

11600

12700

3570

650

-300

5000

4200

2500

-5800

9000

-3300

14300

13100

24100

9700

 

Also the text box “Index of first residue” should for this example be set to 44. The user should choose a probabilistic model, that matches (or approximates) the used NMIA concentration (in this case 13.5mM; available options are 9mM, 13.5mM and 18mM NMIA). Note that the server can handle missing experimental values; they should be indicated as “NA”, as shown in above example for residue 52. The server interpolates between the closest values that are available (in above example the NA value will be replaced by (-600+(-1000))/2=-800). The leaving the checkbox at the bottom of the submission page unchecked, leads to the default behavior of using a 90-nucleotide sliding-window approach for RNA SHAPE data sets corresponding to more than 90 nucleotides. The reason for this approach is that the method is based on training data sets corresponding to at most 88 nucleotides. The SHAPE value normalization can be started with the button “Analyse”.

 

Output Format

After a short while (up to several minutes), a heading “Results files” appears, as well as a link “Results for submission <YOUR_JOB_ID_NUMBER>. Following on that link, a list of result files are presented to the user.

 

The returned output form contains for each residue a posterior probability of cWW-type base pairing.

This file contains residue indices and normalized SHAPE data values.

Additionally, the following files are returned:

1.         params_<YOUR_JOB_ID>.txt
This file contains the input parameters submitted by the user.

2.         basepair_prob.pdf
This file contains a graphical representation of the per-residue posterior probabilities of base pairing.

3.         normalized_signal.pdf
This file contains a graphical representation of the normalized SHAPE data values. It is computed as one minus the posterior probabilities presented in the file basepair_prob.pdf

4.         rnashape_processed.txt
The file rnashape_processed.txt contains in tabular text format the data values used for the graphical representations. The column BP corresponds to the posterior probability of base pairing, the column M1BP corresponds to the normalized SHAPE signal.

 

Reference

Correlating SHAPE signatures with three-dimensional RNA structures.
Bindewald E, Wendeler M, Legiewicz M, Bona MK, Wang Y, Pritt MJ, Le Grice SF, Shapiro BA.: RNA (2011)